Hierarchical service chain orchestration for multi-cloud environments enabled by deep reinforcement learning

With the rapid adoption of multi-cloud platforms, dynamic orchestration of service function chains faces coupled challenges. This study proposes a hierarchical service chain orchestration for multi-cloud environments, which decomposes end-to-end quality of service into real-time, economic, and stability objectives. This model uses hierarchical orchestration structure and integrates Actor-Critic networks with Graph Isomorphism Network and attention mechanisms. The bandwidth allocation layer optimizes bandwidth coefficients for deadline-sensitive tasks, while the server selection layer processes dynamic effective subgraphs for cost-aware server deployment. By Filtering invalid resource nodes via multidimensional constraints to reduce computational load, and prioritizes user requests using a weighted scoring model. Validated on a real-world cloud platform, our method achieves remarkable performance improvements across different load scenarios. Compared with the optimal baseline, it attains a comprehensive score enhancement of 2.56\%, 3.44\%, and 16.8\% under idle, normal, and stress states, respectively. Meanwhile, it achieves a reduction of 0.25\%, 0.35\%, and 0.45\% in the server overload percentage corresponding to the aforementioned scenarios. The model enables flexible and efficient service chain orchestration in multi-cloud.